Coase again

By Kathy G.

First of all, to blogger NutellaonToast who characterized the exchange between Megan McArdle and me as a "catfight": cut out the dumbass sexist bullshit right now, please. It's insulting, dehumanizing, and very, very cheap. And it reveals your anti-McArdle crusade in the worse possible light, because it really does make you look like a gender-obsessed freak who's flipping out at the idea of an outspoken woman who publicly advocates opinions very different from your own. You'd be far more credible if you spare us the sleazy sexism and argue your case against McArdle strictly on the merits. Trust me, fella, it's not that hard.

Secondly, contrary to what NutellaonToast says, neither I nor Megan McArdle are economists (and McArdle, who also describes me as an economist, is also mistaken in this respect). McArdle has been quite upfront about her non-economist status, but people keep making that mistake. And though I frequently blog about economic topics, I'm not an economist, either. I'm a Ph.D. student in public policy and I've taken grad-level econ courses both in the public policy school and the econ department, but as of now the only titles I can rightfully claim are "student" and "blogger."

Now back to McArdle. In her response to me she only digs herself in deeper, I'm afraid. The first time I responded to one of her posts, by taking her to task about her mischaracterization of monopsony, she refused me the common courtesy of linking to me in her reply to that response. This time, she addresses me by name, but this is the link she gives, which, as you can see, goes exactly nowhere. I'd like to give Megan the benefit of the doubt regarding her intellectual honesty and sense of fair play, but lapses like that don't inspire confidence. (Also the first sentence of her post has words missing, which makes it impossible to discern what she means there, but that's another matter).

Here's what Megan has to say about my Coase post:

Kathy G. is further confused by the concept of "preference maximization". Often when I speak to non-economists, I try to explain things in the terms that they are most likely to understand, rather than the terms that appeared in my Micro textbook. I find this helpful, because almost none of them have read my Micro textbook. Luckily, the person I was talking to was not among those who failed to understand what "preference maximization" meant. Indeed, so far as I can tell, Kathy G. is an army of one in this respect. But confess I have not done a comprehensive research survey; I am relying entirely on the absence of confused emails, comments, or blog posts from anyone besides Kathy G.

Here's the thing, Megan: I was confused by the concept of "preference maximization" because it's a concept you pulled out of your own ass. It is a term that I've never heard in class and that appears nowhere in my extensive collection of economics books (believe me, I looked). Then you make things worse by explaining that "preference maximization" is:

the idea is that absent transaction costs, no matter who you endow with the initial bargaining right, the person with the strongest preference will end up with that preference satisfied.

No no no no no no no. What the Coase theorem says is that, when transaction costs are zero and property rights are well-defined, the outcome will be efficient. And what most economists mean by that is, that no party will be worse off, and at least one party will be better off. That means something very different than that "the person with the strongest preference will end up with that preference satisfied."

Now, if we're translating preferences into dollar terms, efficiency means that the good, or the property right, will go to the party with the highest willingness to pay. But no way in hell does "highest willingness to pay" mean "strongest preference." A poor person living near a toxic waste dump may have a very strong preference indeed to see that waste dump removed. But even if transaction costs are zero and property rights are well-defined, that doesn't mean she'll prevail in a Coase bargaining scenario. If she has less money than the polluter, her preference, no matter how strong, will not be realized. All the Coase theorem says is that, after bargaining, she won't end up worse off. But if her resources are less than her opponent's, her preferences would not be satisfied.

Indeed, someone here is confused, but to quote Bob Dylan, it ain't me, babe.

Megan continues:

Kathy G additionally says I should have specified that Coasean bargaining is impractical in the absent of clear property rights. I thought that went without saying, since the discussion revolved around who had a clear property right.

Okay, Megan, let me get this straight. First you're telling us that in inserting your made-up term "preference maximization" in place of "efficiency," you dumbed down the Coase theorem for the benefit of your readers, who are non-economists. Now you say that you didn't mention the "well-defined property rights" qualifier because you assumed that same audience already understood it.

This, I believe, is what is called a contradiction.

In case you can't tell, I take economics really, really seriously -- to the point of seeming like a pedant and a pain in the ass to a lot of people. But that's because economic ideas have a profound effect on public policy, and therefore a far-reaching effect on life as we know it. And because of that, those of us writing for the general public on this subject bear a heavy responsibility to get it right. This, admittedly, is not an easy thing to pull off. Economics is a deep and extremely complex subject. It takes people years of study to master even one small facet of it. Its ideas are often so complicated that the only language precise enough to express them accurately is mathematics. And translating mathematical abstractions into plain English for a nonspecialist audience is not a simple task.

Blogging about economics is an even more difficult task, because in the day-in, day-out nature of the thing, you're bound to screw up. I certainly have. But I try to minimize the screw-ups by being very careful whenever I write about economic topics.

I've read you for a while, Megan, and I've noticed that when you address the topic of economics, you make fundamental errors on a fairly regular basis. Just in the past couple of months, you've been busted for your writings on the Coase theorem, on monopsony (which I link to above), and (by Henry Farrell) on revealed preference. It's hard for me to determine whether these mistakes are a result of careless and lazy writing, or of not having a very firm grasp of the subject. I think it's more a matter of the latter, but who knows?

But the point is, your readers deserve better. You, Megan, are in a privileged position. You're not some random person blogging out of their mother's basement -- on the contrary, you've got a sweet gig as a blogger for the Atlantic Monthly, a magazine with a history as illustrious as any in America, publisher of Emerson, Thoreau, Dickinson, Whitman, Twain, Hemingway, Nabokov, Bellow, and other greats. I don't think it's asking too much that the Atlantic Monthly hire a blogger who a) knows what she's talking about and b) takes care, in her writing, to get it right.

UPDATE: As Atrios points out, "while the term 'preference maximization' is occasionally used, it's a rather bad and misleading substitute for "utility maximization." I'd never heard the term before McArdle used it, so I did some digging around. It didn't appear in my economics books but if you do a Google search, you'll see that the term does come up. Only apparently it doesn't mean "utility maximization," which is what McArdle seems to imply it means. Though who knows.

Comments

[Walter Crockett]

If you hover over Megan's link, you can see that she messed it up and included some of her text in it, along with an additional link. Viewing the page source exposes most of the missing words:
"Kathy G [here's the begining of her link to your page] that I failed to point out that perfect Coasean conditions never hold in the real world. This is absolutely correct. There are also [here she put another link, to http://homepages.wmich.edu/~baldner/black.htm"] no universes composed entirely of two identical spheres. Additionally, the inalienable rights with which all men are self-evidently endowed are frequently, in practice, alienated. I regret the omissions."

[donna]

Well, there do seem to be some economic articles on preference maximization, whatever it is.

I mean, come on, Megan isn't capable of making things like that up. Even if she does have a cushy gig.

[r€nato]

I know it's important to debunk wingnut propaganda - especially to try to keep it from getting out there as 'common wisdom - but WOW this is way into the weeds...

[Max Renn]

I know it's important to debunk wingnut propaganda - especially to try to keep it from getting out there as 'common wisdom - but WOW this is way into the weeds...

Posted by: r€nato | April 26, 2008 at 12:26 PM

But since Meagan's whole project is to legitimize her wacky Randroid ideas in ways that sound like sound theory to the uninitiated, she needs to be busted whenever she tries this nonsense.

Go Kathy!

[knzn]

I think the problem is that there is no a priori correct way to compare strength of preferences across individuals. Until someone specifies a way to do such a comparison, the statement that "the person with the strongest preference will end up with that preference satisfied" is not so much wrong as meaningless. The implicit means of comparison Megan uses seems to be to index strength of preferences by how much someone is willing to pay. The problem with that means is that it depends on endowment as well as preferences.

Of course, one could imagine a social welfare function in which the weights are chosen according to individuals' endowments, but from a normative point of view, such a social welfare function doesn't seem very reasonable. But if one did choose such a social welfare function, and if (as does seem, in general, reasonable to me) one were to define relative strength of preferences according to the impact on the social welfare function, then it would actually be true that a poor person living near a toxic waste dump has only a weak preference for having the dump removed.

[Chris]

Interpersonal comparisons of utility happen constantly. To speak of politics, or indeed of public policy, without admitting this obvious fact, is tantamount to deliberate intellectual dishonesty.

The Coase Theorem is all well and good, but I would offer that in the vast majority of actual cases where legislatures/regulators are involved, the decision certainly is not one which results (or is even intended to result) in a Pareto improvement (much less the achievement of Pareto optimality).

I have no idea who this McArdle person is, but in reading several of her posts at random just now I would say she is one of those people who behave as though their vocabulary is 20% larger than it truly is, and who have the *manner* of educated condescension down, but cannot back it. She seems, in short, to be a poseur.

(I now see she is a lowly MBA (her words). Mystery solved.)

[Echidne]

Congratulations on the great blog, Kathy.

I don't recall ever needing to use the term "preference maximization" in an Econ101 course. "Revealed preference", yes.

Neither do I recall that way of reading the Coase Theorem, and knzn is correct that the particular outcome could only be concluded if all participants started with equal resources.

What is important about the Coase Theorem is the idea that it doesn't matter from efficiency pov if we give the polluters the right to pollute or if we give the victims of pollution the right not to be polluted; as long as there are no transaction costs and as long as both parties can negotiate with each other the final outcome will be efficient.

Sadly, this tells us nothing about the fairness of the process. Just its efficiency. More sadly, there are always transaction costs in the real world and it is often the case that negotiation of the type the theorem requires is impossible. For instance, the polluters may not be known or they may dwell in another country.

Also, they may be much more powerful than the victims of the pollution, but economics is not very good with the concept of power.

[Dave M]

Far from in the weeds, the Coase theorem is one of the central insights of modern law and economics, the foundation upon which they've built a teetering, Escheriffic house of bullshit sophistry. Coase's own work I always liked -- he seemed to have an appropriate understanding of some of his work's relationship to reality, and viewed is L&E "acolytes" with skepticism.

[JayAckroyd]

In the interest of true pedantry, I'd just like to note what the Arrow-Debreu model says about preferences, as told by Marcel Richter.

Under some remarkably weak preference assumptions (identical bundles are equally preferred, you can always choose between two bundles, and an assumption considerably weaker* than transitivity [if a>b and b>c then a>c] that I cannot recall), it can be shown (as I did at least once on the general exam) that these preferences can be represented by a twice differentiable utility function. That is, the same preference ordering arises under such a utility function as arises under a preference ordering that conforms to those assumptions.

This is very exciting (really) because it means that you can generate testable hypotheses from the utility function, and confirm or deny that consumers behave as if their preference ordering reflects those weak conditions.

That means we don't have to, as many people seem think we do, assume that people have utility functions. We can show that they behave as if they do, because we can test whether or not consumers behave in ways that are implied by manipulation of the utility function. We can check and see stuff about price elasticities, downward sloping demand curves, etc are actually reflected in the real world. And if they are, prove that people exhibit this "regular" preference ordering.

In other news, rooting through my books looking for that other assumption, I do indeed find a reference to "preference maximization." In the North Holland Handbook of Mathematical Economics vol II, Barten and Boehm make such a reference in chapter 9, on Consumer Theory.

-----
*It is embarrassing not to remember this because it was a big deal to Richter, rightfully so IMO. It's not at all clear that if you prefer say, a banana to a cheese sandwich, and a ham sandwich to a bag of potato chips that you will always prefer a banana to a bag of potato chips.

[JayAckroyd]

make those sandwiches the same in the above illustration, please.

[CMike]

Let me get this straight. You looked through your text books for the words "preference maximization" and tried to recall whether you had heard the term discussed in class. Then you sat at your computer and, without googling the term, blogged, "I was confused by the concept of 'preference maximization' because it's a concept you pulled out of your own ass."

Sure McArdle did a bad job with the concept. However, I have to agree with her that "preference maximization" is not a perplexing term nor is it, for that matter, an obscure one.

[liberal]

The Coase Theorem is false, because it assumes willingess to pay is equal to willingness to accept.

[hellblazer]

I just remember the days when McArdle used to redirect incoming links from critical posts. The blogger's equivalent of putting one's fingers in one's ears and saying "naa naa naa I can't hear you".

Why she has her position at Atlantic, I can't say. However, it was enough to cancel my subscription to their magazine.

[JayAckroyd]

Cmike,

It actually is a perplexing term. I was surprised to find it. A preference ordering has no value. It's hard to understand what it means to "maximize" an ordering. The reason one seeks to show that preference ordering can be investigated by creating a utility function that corresponds to that preference ordering is that it is a well formed expression to say "maximize a utility function."

I started reading Megan during the flap with Glenn Greenwald. But I stopped when she, apparently purposely, misrepresented a NYT Mag article about wages in the Rust Belt. I can't read blogs that I have to click through to confirm everything.

[Kathy G.]

CMike, I did indeed Google the term after reading McArdle's first post, and it came up, but it's a (relatively) rarely used term in economics, which in any case, does not mean "utility maximization," which is what McArdle implied it meant. More importantly, the Coase theorem predicts nothing about either utility maximization or preference maximization -- it only makes a prediction about efficiency.

I very seriously doubt that McArdle had ever heard the term "preference maximization" before she used it in that context. Her usage of it gave no evidence that she knew what it meant. And in her response to me, she more or less admits it's a term she herself uses to mean "the person with the strongest preference will have that preference satisfied." Which is also false.

I gotta give McArdle credit -- in just a few short sentences she managed to cram in an astonishing quantity of bullshit,

[brad]

Hmmm, I want to be careful not to knock my co-blogger, who can be quite funny and meant no disrespect, but I can't say I entirely disagree. I've emailed Kathy and Nutella privately, so let me just say here that the thin veneer of misogyny is meant to barely cover a strong core of respect. There is no diminution intended.

[lemuel pitkin]

What the Coase theorem says is that, when transaction costs are zero and property rights are well-defined, the outcome will be efficient. And what most economists mean by that is, that no party will be worse off, and at least one party will be better off.

No. That is not what the Coase Theorem says.

What the Coase Theorem says is that if property rights can be traded freely, the ultimate allocation of a particular right is invariant to its initial allocation.

For example, my neighbor is noisy, I'd like her to be quiet. It doesn't matter whether she has the right to be noisy, which i must pay her not to exercise, or I have the right to peace and quiet, which she must pay me to violate. Assuming there is no obstacle to us making a binding contract on noise levels, she will end up making exactly the same amount of noise in both cases.

This is a significantly stronger result than your version, and it has important policy implications, e.g. it is one way of seeing the formal equivalence of carbon permits and a carbon tax.

I like your work a lot, and Megan is a blithering idiot, but you've got this one wrong.

[lemuel pitkin]

First paragraph is a quote from Kathy G.

(I forgot html doesn't work here. Why is that, by the way?)

[Rob]

Except thats their is another big assumption that is often ignored, and that is of full information.

[Frankly, my dear, ...]

"I'd like to give Megan the benefit of the doubt ..."

What doubt?

[CMike]

In Wikipedia at Utility under Utility Function it reads

"While preferences are the conventional foundation of microeconomics, it is convenient to represent preferences with a utility function and reason indirectly about preferences with utility functions."

You can link to "Preferences" where, under the heading "Preference in Economics," it reads:

"In microeconomics, preferences of consumers and other entities are modelled with preference relations."

Yeah, I understand there are all types of considerations about cardinal, ordinal, marginal value, but to say you are stumped as to the meaning?

Over at Google Books I see that some Canadian economist, C. René Dominique, wrote:

"Preferences are then mapped into a utility index, and economists claim that this is utility maximization rather than preference maximization, which dates back to Frisch's (1926) initial axiomatization. Actually, preference maximization was thought to be the more basic concept of consumer behavior, but utility maximization was perceived as more convenient to handle..."

Now I don't know who Frisch was or who Dominique is. But gee whiz, the use of the term "preference maximization" was not what was wrong the McArdle's post.

[CMike]

Whoops,

I should have indicated that my 6:22 PM comment was in response to JayAckroyd who I see elsewhere. I'm not trying to pester Kathy G. I appreciate the direct response I got from our host.

[JayAckroyd]

Kathy,

I did indeed Google the term after reading McArdle's first post, and it came up, but it's a (relatively) rarely used term in economics, which in any case, does not mean "utility maximization,"

It pretty clearly does mean that in the review article in the North Holland Handbook.

It also has a reference to the Ragnar Frisch 1926 article.

I don't remember anyone using the term. But I had read the chapter sometime long ago.

[Kathy G.]

Lemuel, there are a number of different variations of the Coase theorem out there. I chose the variant that makes the weakest claims, since that's the one that (I think) is the least controversial. And that's the one that says the ultimate allocation will be efficient.

I steered clear of the version that says the ultimate allocation will be invariant, because that's a much stronger claim. The version I learned and the one that is most commonly discussed in the literature I've seen makes the efficiency claim, not the invariance claim.

[Matt Weiner]

Lemuel: I'm not an economist or close to one, so I haven't encountered the Coase theorem in any sort of class. But I see that the Wikipedia page (yeah, I know) on the Coase theorem says that Coase initially gave a loose formulation, and there are (at least) two sharpenings of it. The less controversial efficiency theorem is the one that Kathy put forward; the version you describe seems to be the invariance version, which is apparently more controversial.

For what it's worth, the page at the U of C law school (where Coase is) devoted to the theorem gives the efficiency version:

"The Coase Theorem can be simply stated: in a world where there are no transaction costs, an efficient outcome will occur regardless of the initial allocation of property rights."

And the invariance version seems obviously too strong without the addition of some other assumptions. If I don't have enough money to pay my neighbor not to be noisy, and she doesn't have enough to pay me to let her be noisy, then whether she's noisy will depend on the initial allocation of rights. I've seen some versions of the Coase theorem that say that it depends on assuming there are no wealth effects; is that what's going on in the invariance version?